Notes and Tables, FY 1994

Technical Notes


Scope of the SurveyTOP

Data for the National Science Foundation's (NSF's) fiscal year (FY) 1994 report on research and development (R&D) expenditures were collected from 500 institutions of higher education in the United States and Outlying Areas. These institutions were selected from the universe of 681 schools that have doctoral programs in science and engineering (S&E), are historically black colleges or universities that expend any amount of separately budgeted R&D in S&E, and/or are other institutions that expend at least $50,000 in separately budgeted R&D in S&E.

In addition, the survey includes 18 federally funded research and development centers (FFRDCs). To qualify, an FFRDC must be engaged in basic or applied research, development, or management of R&D activities, and the results of these activities must be directly monitored by the Federal Government--usually a single agency--in a relationship expected to be maintained on a long-term basis. The center must be operated, managed, and administered by either a university or consortium of universities as an autonomous organization or as an identifiable separate operating unit of its parent institution. Finally, 70 percent or more of the center's financial support must be received from the Federal Government.

Although the same survey form (NSF Form 411) is used to collect data from both academic institutions and FFRDCs, the resulting data are presented separately in this report. The survey population was reviewed prior to mailing the questionnaires to ensure that each institutional classification was accurate. Characteristics of the schools were reviewed before and during the course of the survey to determine if changes had occurred (i.e., in highest degree granted or in terms of school openings, closings, or mergers).

Survey Instrument TOP

Most major R&D performers have incorporated into their recordkeeping systems the data that are essential to complete this survey, thereby ensuring a consistent format from one year to the next. Such consistency yields the most useful statistics for time series. As a rule, information to complete this questionnaire is found within the institutions' year-end accounting records.

The survey questionnaire consists of three main items:

Item 1 is a request that institutions report their total current expenditures for separately budgeted science and engineering R&D for all activities specifically organized to produce research outcomes and commissioned by an agency either external to the institution or separately budgeted by an organizational unit, i.e., research centers, within the institution by source of funds. In addition, schools are asked to provide the percentage of the total and the percentage of the federally financed expenditures that are considered basic research. Included also are research funds for which an outside organization, educational or other, is a subrecipient. Care should be observed when interpreting data on source of funds; for example, industry R&D support is limited to grants and contracts for R&D activities from profitmaking organizations, and the total reported excludes research funded through unrestricted accounts and from corporate foundations, endowments, and fellowships to students. An increasing number of institutions have linkages with industry and foundations via subcontracts, thus complicating the identification of funding source. In addition, institutional policy may determine whether unrestricted State support is reported as State or as institutional funding.

Item 2 is a request for total and federally financed current fund expenditures for separately budgeted R&D activities by detailed S&E fields. Major fields remain unchanged from the FY 1993 questionnaire. When interpreting these data at the detailed discipline level, users should keep in mind that there is considerable interdisciplinary activity and/or overlap among subdisciplines.

Item 3 is a request for the portions of total and federally financed expenditures reported in items 1 and 2 that were used for the purchase of research equipment out of current funds. This portion includes all research equipment purchased under sponsored research project awards and disbursed in the same detailed disciplines as in item 2. These data are of special interest to Federal and institutional policymakers in determining current funding levels for scientific research instrumentation.

Response Rate TOP

The FY 1994 survey questionnaires were mailed in October 1994. Every effort was made to maintain close contact with respondents in order to preserve both consistency and continuity in the resultant data. Questionnaires were carefully examined for completeness upon receipt. Computerized facsimiles of the survey data were then prepared for each institution, comparing the current and 2 prior years' data and noting any substantive disparities. These facsimiles were mailed to the respondents so that they could provide revisions before final processing and tabulation of the data.

Respondents were asked to explain significant discrepancies between current and prior years' reporting patterns previously verified as correct. They were encouraged to correct prior years' data if anomalies were identified. When updated or amended figures covering past years were submitted, trend data were correspondingly changed by NSF. Similarly, if a respondent institution underwent an organizational change, such as a merger, NSF incorporated the effects of such changes into prior years' data.

By the survey closing date in mid-June, forms had been received from 498 universities and colleges out of the academic sample population of 500, resulting in a 99.6-percent response rate. Responses were received from 99.7 percent of all doctorate-granting institutions, where 97.8 percent of the R&D expenditures in S&E fields was disbursed. Also, forms were received from all 18 FFRDCs. Tables A-1a and A-1b display a detailed breakdown of the response rates by highest degree granted and by sampling stratum.

National Total and Imputation TOP

To provide a national estimate for all universities and colleges performing R&D in FY 1994, it was necessary to implement three statistical procedures. First, data were estimated by "imputation" for less than 1 percent of the sample population that had not responded by the closing date of the survey. Imputation has been used consistently since FY 1976. Second, data were also imputed for universities and colleges that submitted only partial responses. The imputed total, prior to weighting, was $5 million. Third, the sample total was weighted to compensate for those universities and colleges that were in the survey universe but not in the survey sample. This process led to an inflation of $204 million in the national total of R&D expenditures at universities and colleges for FY 1994, resulting in a $21.1 billion total, as shown in table A-2. (The imputed total was inflated to $7 million, as noted in table A-2.)

Tables A-3a and A-3b present breakdowns of both the imputed amounts and the amount of the weighted inflator by broad S&E field. The dollar amount imputed is displayed along with the percentage it represents of the national estimate for universities and colleges in a particular field. Also given is the amount of the weighted inflator for that field. The amount imputed is similarly broken down by source of funds in table A-4.

A significant number of surveyed institutions have been responding only intermittently in past years, providing data one year, not responding for one or more subsequent years, and then providing data again. For the years in which no response was received, data have been imputed as previously described. Although the imputation algorithm accurately reflects national trends, it cannot account for reporting anomalies at individual institutions. For this reason a separate backcasting of prior years' data was performed, following current-year imputation.

For each institution, formerly imputed key variables for items 1 to 3 were compared with subsequent submissions to determine whether the imputed data accurately represent the growth patterns shown by reported data. Re-estimation was applied when the imputed data were not consistent with reported data. If data were reported for fiscal years 1991 and 1994 but not for the intervening years, for example, the difference between the reported figures for each item total was calculated and evenly distributed across the intervening years (1992-93). The new figures were spread across disciplines (items 2 and 3) or sources of support (item 1) on the basis of the most recent reporting pattern. A clean facsimile was generated for each of the institutions undergoing these procedures and returned to the school for comment. These procedures resulted in much more consistent reporting trends for individual institutions but had little effect upon aggregate figures reflecting national totals.

Data Anomalies TOP

Aggregate academic expenditure data are generally consistent from year to year, although data for individual institutions may vary considerably. Data anomalies may reflect true increases or decreases in expenditures or may be the result of changes in reporting methodology.

Highest-Degree-Granted Tables TOP

Several longitudinal tables display data for institutions whose highest S&E degree granted is at the doctoral level. In tables produced prior to FY 1992, it would have been difficult to identify whether changes in yearly R&D expenditures were caused by changes in expenditure levels or in the number of doctorate-granting institutions. In order to maintain a consistent group of institutions across all years, the highest-degree-granted status for each institution is based on the highest degree granted in the most recent year, FY 1994.

Sampling, Weighting, and Standard Errors of the EstimatesTOP

Full population surveys are conducted by NSF every 5 years. During intervening years, a sample of institutions is drawn and surveyed. Only universities and colleges are included in the sampling frame; FFRDCs are always surveyed. Since a full population survey was conducted in FY 1993, a new sample of institutions was drawn for the FY 1994 survey. This sample will be maintained and resurveyed for the next 3 years until the next full population survey in FY 1998. Universities and colleges have been divided into the following four sampling strata (three certainty strata and one probability stratum):

  1. A certainty stratum of doctorate-granting institutions. This stratum contains 316 universities and colleges that have doctoral programs in S&E. Excluded from this stratum are all doctorate-granting historically black colleges and universities (HBCUs) and university-system campuses.

  2. A certainty stratum of HBCUs. This stratum includes all 65 HBCUs (including those that are doctorate-granting institutions or university-system campuses).

  3. A certainty stratum of university-system campuses. This stratum includes 56 "university-system campuses"; each sample entry is the aggregation of all campuses that make up a university system. Note that no HBCUs are included in this stratum even if they are university-system campuses.

  4. A probability stratum of institutions that grant degrees at the master's level or below. This stratum includes 63 master's or bachelor's degree-granting institutions and institutions that offer no S&E degrees at any degree level, out of a survey stratum universe of 244. Note that none of the universities or colleges included in this stratum is either an HBCU or a university-system campus. This stratum had a sampling ratio of 26 percent.
The data in this report are weighted to represent national-level R&D expenditures for universities and colleges. The sample data, after imputation, were inflated to produce universe estimates by weighting the individual questionnaire data values by the inverse of the sampling ratio. Thus, in aggregating data for institutions from the probability stratum for tabulation purposes, each datum value was weighted by the inverse of the sampling ratio.

Estimates derived for institutions in the probability stratum were based on a sample, and the relative standard error (coefficient of variation) of an estimate was then obtained by dividing the standard error by the estimate itself, expressed as a percentage of the estimate.

The standard errors and coefficients of variation for each major S&E field are shown in table A-5. For example, for total academic R&D expenditures of $21.1 billion, the standard error of the estimate is $98.7 million at the 95-percent confidence level, with a coefficient of variation of ±0.5 percent. Similarly, for the estimate of $12.7 billion in federally financed expenditures, the 95-percent confidence limits are ±$54.3 million, with a coefficient of variation of ±0.4 percent.

Data Availability TOP

Data published in this report are also available on diskette and on the World Wide Web. Single-year or multi-year data files are available with data for FY 1975 through FY 1994. Information on file formats and the years for which they are available, with instructions for downloading, can be found on the World Wide Web at, or in the current data user guide, Guide to the Data Files from the National Science Foundation's Annual Surveys of Academic Science and Engineering. To obtain the guide in hard copy form, contact M. Marge Machen.

Selected data items for individual doctorate-granting and master's-granting institutions are available on computer-generated institutional profiles. These profiles cover data from this survey as well as data collected in NSF's other academic S&E surveys: the Survey of Graduate Students and Postdoctorates in Science and Engineering (graduate student survey) and the Survey of Federal Support to Universities, Colleges, and Nonprofit Institutions (Federal support survey). Institutional profiles are also available on the World Wide Web (, or can be ordered in hard copy form through Mr. Richard Bennof. He can be reached at--

Research and Development Statistics Program
National Science Foundation
4201 Wilson Boulevard, Suite 965
Arlington, VA 22230

Telephone: (703) 306-1772
Institutional researchers can obtain data from several academic S&E resources through the Computer-Aided Science Policy Analysis and Research (CASPAR) database system, which is an easy-to-use tool for the retrieval and analysis of statistical data on academic S&E resources. CASPAR provides an extensive and growing data library with multi-year statistics on the state of higher education in general and on academic S&E resources specifically. This data library is based on a set of standard institutional and field-of-science definitions across the multiple sources used to develop the database. The CASPAR program includes built-in help capabilities to facilitate the use and interpretation of the data.

The latest version of CASPAR can now be accessed by downloading from the FTP server either directly ( in the /pub/caspar directory) or via the World Wide Web ( A CD-ROM compact disk is also available. For information, contact M. Marge Machen.

CASPAR data are drawn from a number of sources. All data are available on the individual institutional, State, and national levels. Longitudinal data from surveys of universities and colleges conducted by the NSF Division of Science Resources Studies include this R&D expenditures survey, the Federal support survey, and the graduate student survey. Data from surveys of universities and colleges conducted by the National Center for Education Statistics include Earned Degrees; Opening Fall Enrollment; Faculty Salaries, Tenure, and Fringe Benefits; and Financial Statistics. Data from other sources include the National Research Council Doctorate Program Ratings.